I recently converted a machine over to linux too. I am some what okay with the fact that linux gets less points per result because it work is done faster. However, I am mad that linux gets less points per hour of work also (it is particular bad if you have the 32 bit version). Now if you are doing some of the zero redundancy projects, then you tend to get about the same per hour of work you do.

That being said, I have to admit logically, I would have to say if linux does work faster then it should rewarded more points. However at the same time, how can we determine the correct amount of points to reward? We can't say linux will always process work 50% faster (made up a number), so give it twice the amount of points per hour (if windows takes 2 hrs, and linux takes 1, then you need to double the points). The reason why we can't is because the performance difference between linux and windows varies for every project. HCC maybe 40% faster but c4cw may only be 30% faster (made up these numbers, didn't want to go search for my post where I did a comparison).
If the WU from a done linux machine can be compared to WU done on a windows machine for validation then we can do comparison there. However, most windows user will even more angry because sometimes they will get 100pts for a WU (when compare with another windows machine) but sometimes they will get only 50 pts (when compare with a linux machine).
Now for zero redundancy, they can actually improve it here. The little performance test that they run to determine pts per hour is currently probably a standardize test (guessing here). They can make it more specific, the test should be a small snippet of a normal WU for the project. This way they can a more accurate feel of how the machine will perform for their WU

having each WU get a set number of points (based on the project) can't always be done. This is because for some projects WUs vary in length. For example, for the HFCC project, some of them takes 4 hrs, some of them take more then 10. This is on the same system, so it wouldn't be good to award a fix point per project.
However, it could work for c4cw. They seem to have 3 types of WUs and for each type the amount of time are very similar.

Lets be honest though, there is no 100% perfect point system. People will always have soemthing to complain about.

My HCC crunching is entirely done under Linux. I can definitely see how low the points are.

At this point TeamHedy is pretty much us, so looking at the Team Competition stats for HCC points vs HCC results shows the underscoring.
While I can't guarantee the other teams are using Windows, the odds are they are. 178 per result is way below average.

I care about getting things done, so I think the points structure should be guiding people to the most efficient hardware and SOFTWARE.



HCC Points
5 Vulture Central III 85,057
6 MyOnlineTeam 77,959
7 BRASIL - BRAZIL@GRID 77,213
8 UK 67,170
9 TeamHedy 53,458

HCC Results
4 Vulture Central III 311
5 TeamHedy 300
6 MyOnlineTeam 292
7 BRASIL - BRAZIL@GRID 274
8 UK 235

Points Per Result Average
Vulture Central III 273
MyOnlineTeam 266
BRASIL - BRAZIL@GRID 281
UK 285
TeamHedy 178

When the performance improvement is great like for HCC or C4CW, yes the daily points maybe slightly better under Linux than under Windows, but only slightly. But when the performance benefit is smaller like for FAAH or HFCC the daily points of a Linux device can show a loss versus Windows, although more work has been produced.

To complement my previous table here is another table based on numbers in the previous one. I think numbers on the last line will not need additional comments:

Daily production       WUs     Points
Q6600 XP 32-bit       54,01   2191,39
Q6600 Ubuntu 64-bit   88,15   2343,03
Ubuntu/Windows       +63,21%    +6,92%